I believe when they said "self-published, unknown" journals, the key word was unknown rather than self-published. They're not talking about journals self-published by established scholarly societies and so forth, who are "known." Nature (or other known journals) come with an established reputation (deserved or not) for publishing high quality work; an unknown journal (i.e. one with no established reputation one way or the other) by definition cannot provide this.

In recent years, journals that claim to have a peer review process but which actually offer dubious or no review or quality control have proliferated (some are even published by companies like Elsevier). Having no established reputation doesn't prove that a journal is bad, but having an established reputation is a heuristic shortcut for evaluating whether it is good (at least in the "no one was ever fired for buying IBM" sense). Hiring committees use this heuristic to save time when filtering candidates.